Contralesional Homotopic Functional Plasticity in Patients with Temporal Glioma

CLINICAL ARTICLE J Neurosurg 134:417–425, 2021

Contralesional homotopic functional plasticity in patients with temporal glioma

*Dongming Liu, MS,1 Jiu Chen, MD,2,3 Xinhua Hu, MD,1,3 Guanjie Hu, MS,1 Yong Liu, MD,1 Kun Yang, MD,1 Chaoyong Xiao, MS,3,4 Yuanjie Zou, MS,1,3 and Hongyi Liu, MD1,3

1Department of Neurosurgery, The Affiliated Brain Hospital of Nanjing Medical University, Nanjing, Jiangsu; 2Institute of Neuropsychiatry, The Affiliated Brain Hospital of Nanjing Medical University, Fourth Clinical College of Nanjing Medical University, Nanjing, Jiangsu; 3Institute of Brain Functional Imaging, Nanjing Medical University, Nanjing, Jiangsu; and 4Department of Radiology, The Affiliated Brain Hospital of Nanjing Medical University, Nanjing, Jiangsu, China

OBJECTIVE This study aimed to explore the contralesional homotopic functional plasticity in the brain of patients with unilateral temporal glioma. METHODS Demographic, neurocognitive, and resting-state functional MRI data were collected from 17 patients with temporal glioma (10 in the right lobe and 7 in the left lobe), along with 14 age- and sex-matched healthy controls. The amplitude of low-frequency fluctuation (ALFF) of the contralesional homotopic region and 2 control regions was exam- ined. The region-of-interest–based analysis was used to determine the altered functional connectivity (FC) of the contralesional homotopic region, showing significantly different intrinsic regional brain activity between patients and controls. Partial correlation analysis was conducted to determine the association between the altered neural activity and behavioral characteristics. RESULTS Compared with controls, patients with right temporal glioma exhibited significantly increased ALFF in the contralesional homotopic hippocampus and parahippocampal region. In addition, the intrinsic regional activity in these regions was negatively correlated with the visuospatial score (r = -0.718, p = 0.045). Whole-brain FC analysis revealed significantly increased FC between the left hippocampus and parahippocampal regions and the left inferior temporal gyrus, and decreased FC between the left hippocampus and parahippocampal regions and the left inferior frontal gyrus. No significant changes were found in the 2 control regions. CONCLUSIONS Contralesional homotopic regions are instrumental in the process of neural plasticity and functional compensation observed in patients with unilateral temporal glioma. The observed findings might be used to help preoperative evaluation or rehabilitation of postsurgical patients. https://thejns.org/doi/abs/10.3171/2019.11.JNS191982 KEYWORDS resting-state MRI; cognitive function; temporal glioma; plasticity; contralesional hippocampus; oncology

liomas account for approximately 81% of malig- increasing number of studies have identified different pat- nant brain neoplasms and are the most common terns of lesion-induced cerebral plasticity by functional primary intracranial tumors.26 In addition to com- imaging,6,7,14,17,33 which are related to the location, extent, monG neurological manifestations, multiple (mild) cogni- type, and development of the lesions. Brain plasticity has tive deficits are also frequently observed in glioma pa- been observed in the perilesional, ipsilateral hemispheric, tients.30 Previous studies have indicated that patients with contralateral hemispheric, and cerebellar regions1,13,43 and glioma might experience potential neural adaptation to the manifests as structural rearrangement and/or functional lesions.1,13,43 Faced with a progressive loss of neural tis- compensation.1,16,38 Functional compensation can be clas- sue, the brain may reassign the remaining physiological sified into specific function and functional networks.13,16,20 resources to maintain a compensatory level of function in Functional compensation frequently manifests as al- a socially and cognitively demanding environment.14 An tered blood flow or metabolism at the resting state or

ABBREVIATIONS ALFF = amplitude of low-frequency fluctuation; DSST = Digital Symbol Substitution Test; FC = functional connectivity; fMRI = functional MRI; GRF = gaussian random field; HGG = high-grade glioma; LGG = low-grade glioma; MNI = Montreal Neurological Institute; ROI = region of interest; TFCE-FWE = threshold-free cluster enhancement family-wise error. SUBMITTED July 22, 2019. ACCEPTED November 5, 2019. INCLUDE WHEN CITING Published online January 10, 2020; DOI: 10.3171/2019.11.JNS191982. * D.L., J.C., and X.H. contributed equally to this work and share first authorship.

©AANS 2021, except where prohibited by US copyright law J Neurosurg Volume 134 • February 2021 417

Unauthenticated | Downloaded 09/29/21 07:53 AM UTC Liu et al. during activation and can be tracked via functional im- inclusion criteria were as follows: 1) histopathologically aging to study brain recovery and plasticity in a nonin- confirmed primary glioma (tumor grade was reexamined vasive manner.13 Functional MRI (fMRI) has been used by the pathologist according to the 2016 WHO classifica- to explore the neural plasticity in patients with glioma.11,43 tion of central nervous system tumors); 2) no evidence of The amplitude of spontaneous brain oscillations can be shift of the midline structures (septum pellucidum, cor- measured as the amplitude of low-frequency fluctuations pus callosum, third ventricle) due to peritumoral edema (ALFFs) to investigate the disturbances in the resting-state or mass effect of the lesion as confirmed by structural im- intrinsic regional activity. ALFF is a feasible and reliable ages; 3) no history of head trauma, cerebrovascular dis- indicator of resting-state regional brain activity,41 which is ease, psychological disease, temozolomide chemotherapy, altered in the presence of brain lesions.12,43 ALFF has been or radiotherapy; and 4) absent or slight neurological focal used to measure diverse brain plasticity, including the sen- deficit including aphasia or paresis. The exclusion crite- sorimotor system,23 motor recovery,21 reorganization of ria were as follows: 1) recurrent glioma; 2) multiple lesion cerebro-cerebellar circuits,43 and functional networks.19 foci; 3) history of substance abuse, including tobacco and Altogether, this suggests that ALFF is a reliable tool to alcohol; and 4) inadequate MRI data acquisition and pre- investigate brain disorders and the concomitant functional processing. Based on these criteria, 17 patients (mean age plasticity. 53.59 ± 13.13 years, 11 males and 6 females) were included Previous studies have shown that slow but massive in- in the study. In addition, 14 age- and sex-matched healthy filtration of the insula with glioma masses can induce a controls were recruited from the local community, after marked increase in gray matter volume in the contralat- excluding those with a history of head trauma, severe sys- eral insula. The homotopic reorganization in the macro- temic disease, or psychological disorders. All participants structure of the cortex might be a physiological basis of 1 provided written informed consent, and the study was ap- functional compensation. In addition, empirical studies proved by the Institutional Ethical Committee for Clini- have demonstrated the importance of contralateral ho- cal Research of the Affiliated Brain Hospital of Nanjing motopic areas in maintaining cognitive and sensorimo- Medical University. tor functions.10,37 Despite the critical role of the temporal 22 structure in multiple brain functions, surgical excision Neurocognitive Assessment of the tumor-infiltrated temporal lobe may not result in serious neuropsychological impairments,14 underscoring All patients were evaluated during a standardized clini- functional compensation and neural plasticity in patients cal interview and neurocognitive tests. The details of the with unilateral focal glioma. Furthermore, unilateral or neurocognitive assessment are provided in Supplemental focal glioma growth not only remodels the regional neural Data Methods S.2.3. activity but also alters the large-scale neural network.4,24,43 Reorganization of the resting-state networks has been ob- MRI Data Acquisition served in various pathological conditions involving brain Patients underwent preoperative MRI between 2013 damage, such as stroke and brain tumors.5,20,36 It is, how- and 2015. All MRI images were acquired on a 3.0-T Ve- ever, not completely clear whether the functional reorga- rio scanner (Siemens) at the Department of Radiology, the nization in patients with hemispheric focal glioma is sup- Affiliated Brain Hospital of Nanjing Medical University. ported by modifications in the contralesional homotopic The details of image acquisition parameters are provided region, or if that plays any role in global network integra- in Supplemental Data Methods S.2.3 and in our previously tion. published study.20 To this end, we used ALFF to explore the functional remodeling in patients with temporal glioma and selected Lesion Drawing and Mask Fabricating possible regions for subsequent region-of-interest (ROI)– Original T1-weighted images were normalized to the based functional connectivity (FC) analysis. We hypoth- Montreal Neurological Institute (MNI) Template us- esized that the homotopic temporal lobe supports func- ing SPM12 (https://www.fil.ion.ucl.ac.uk/spm/software/ tional compensation, which can be detected by ALFF and spm12/) with a spatial resolution of 2 × 2 × 2 mm. The out- might reflect the alteration (or at least a trend of the altera- line of the lesion was then traced manually on individual tion) of clinical manifestations in some way. We further 3D T1-weighted images using MRIcron software (https:// hypothesized that the homotopic functional compensation www.mccauslandcenter.sc.edu/crnl/mricron/). The details is not isolated and involves the entire neural network of of image acquisition parameters are provided in Supple- the brain. mental Data Methods S.2.4.

Methods MRI Imaging Data Preprocessing Participants MRI images were preprocessed using Data Process- We reviewed the records of 32 patients who underwent ing and Analysis for Brain Imaging (DPABI; http://www. temporal/insular tumor surgery at the Department of Neu- rfmri.org) toolkit39 and Statistical Parametric Mapping rosurgery, The Affiliated Brain Hospital of Nanjing Medi- 12 (SPM12; http://www.fil.ion.ucl.ac.uk/spm) toolkit in cal University, Jiangsu Province, China. Since the tempo- MATLAB (release 2013b, http://www.mathworks.com/ ral/insular glioma localization is frequent,6,27 the clinical products/matlab/). The details of imaging data prepro- and MRI data of these patients were readily available. The cessing are provided in Supplemental Data Methods S.2.5.

418 J Neurosurg Volume 134 • February 2021

Unauthenticated | Downloaded 09/29/21 07:53 AM UTC Liu et al.

Contralesional Intrinsic Spontaneous Neural Activity sidered statistically significant. Age, sex, and education Analysis (years) were treated as covariates in all between-group ALFF was used to investigate the intrinsic spontaneous statistical analyses. Pearson’s partial correlation analysis neural activity of the contralesional homotopic region at was conducted to analyze any association between the the voxel level since it can reflect changes of neural activ- ALFF or FC values and clinical scores. Any p value < ity effectively on fMRI.9,12 The details of ALFF calcula- 0.05 was considered statistically significant. tion are provided in Supplemental Data Methods S.2.6. The statistical analysis of ALFF was confined to the ho- Results motopic masks of the contralesional hemisphere. Demographic and Clinical Characteristics To verify that the functional compensation was specific to the contralesional homotopic temporal lobe, the ALFF The demographic data and clinical characteristics col- of the temporal lobe and 2 control regions (thalamus and lected from all participants are presented in Table 1 and medial frontal) were also calculated.1 The reason for Table S1. All patients and controls were of Chinese Han choosing these regions in the contralesional hemisphere descent and right-handed according to the Edinburgh is that the thalamus near the midline might be sensitive to Handedness Inventory. In our study, the period between tumor mass effect (midline shift), while the medial frontal the patient’s first onset of subjective symptoms and the region is relatively far from the contralesional homotopic magnitude of resting-state changes did not exceed 24 regions. To assess the effects of the tumor grade on fMRI weeks (all patients: mean 7.76 weeks [SD 7.79 weeks]). Pa- changes, we also performed analysis ignoring side but tients with low-grade glioma (LGG) in the right temporal comparing contralateral neural activity in low-grade ver- lobe tended to have a longer interval than those with high- sus high-grade tumors. The ALFF values of the patients grade glioma (HGG) in the left temporal lobe before the and controls were compared using t-tests, and the results collection of fMRI data (p = 0.047, t = 2.40). According were corrected by the threshold-free cluster enhancement to WHO classification, 8 patients had LGGs (WHO grade family-wise error (TFCE-FWE) correction.34 The stan- I or II) and 9 had HGGs (WHO grade III or IV). In addi- dardized ALFF values of the regions showing significant tion, 7 gliomas were localized to the left temporal lobe (1 included the insula, LGG/HGG = 1/6) and 10 to the right group differences and that of the control regions were also temporal lobe (2 included the insula, LGG/HGG = 7/3). extracted and compared with that of the healthy controls. No significant differences were observed in age, sex, or education of the patients and controls. When evaluating ROI-Based Whole-Brain FC Analysis neurocognitive characteristics, Bonferroni correction ad- To study the reorganizational pattern of the homotopic justments showed that all patients with temporal glioma region from a network perspective, the regions within the exhibited inferior visuospatial, Digital Symbol Substitu- homotopic mask showing significant group differences in tion Test (DSST), mapping, similarity, mathematics, and ALFF were defined as ROIs for the subsequent FC analy- memory performances (all p < 0.05). Patients with right sis.43 The cluster saved was corrected by TFCE-FWE cor- temporal lobe gliomas especially demonstrated worse rection (p < 0.05). The signal time courses of the ROI were performance in visuospatial, DSST, mapping, and math- extracted using the Resting-State fMRI Data Analysis ematics tests (all p < 0.05), but not in memory and similar- Toolkit (REST, http://resting-fmri.sourceforge.net) soft- ity tests (p > 0.05). ware, and voxel-level correlation analysis was performed to generate the FC map in each subject. The whole-brain ALFF Group Differences in the Contralateral Homotopic FC map was calculated using the Functional Connectivity Region Toolkit in REST. The resultant FC maps were transformed In patients with right temporal lobe glioma (n = 10), into z scores by Fisher’s r-to-z transformation to improve the maximum lesion overlap was located within the right normality for the subsequent t-test. Statistical analysis temporal lobe (Fig. 1A). Compared with controls, these of the FC map z scores was confined to the nonlesional patients showed increased ALFF in the left hippocampus whole-brain gray matter mask. and parahippocampal gyrus within the homotopic mask (permutation test, TFCE-FWE corrected p < 0.05). De- Statistical Analysis tails of the cluster are listed in Table 2. In patients with Statistical analysis was performed using IBM SPSS left temporal glioma (n = 7), the maximum lesion over- (version 19.0, IBM Corp.). An independent two-sample t- lap was located within the left temporal and insular lobes test was used to compare the demographic variables and (Supplemental Data Fig. S1A). No significant difference neurocognitive scores. The significance threshold was set in ALFF analysis was found between these patients and to a Bonferroni corrected p value < 0.05 (7 tests). The sex controls. For the other control regions (i.e., the contrale- ratio was compared using the chi-square test, and nonpara- sional thalamus ROI and medial frontal ROI), the ALFF metric permutation tests were used to compare the ALFF values were similar for both patient groups and controls. values between patients and controls34 using the DPABI Compared with controls, the ALFF values extracted from toolkit.3 The TFCE-FWE–corrected cluster p value < 0.05 the left hippocampus and parahippocampal gyrus were was considered statistically significant. The permutation superior in patients with right temporal lobe glioma for times were set at 1000 tests. Gaussian random field (GRF) the temporal lobe (t = 5.12, p < 0.001), but not for the con- correction was used for FC analysis, and voxel-level p tralesional thalamus (t = 0.45, p = 0.65) and the thalamus < 0.0005 and corrected cluster-level p < 0.05 were con- (t = 1.16, p = 0.26; Fig. 1D), confirming that the intrinsic

J Neurosurg Volume 134 • February 2021 419

Unauthenticated | Downloaded 09/29/21 07:53 AM UTC Liu et al.

TABLE 1. Demographics and cognitive measures of patients with temporal glioma and controls Items HCs (n = 14) All Patients (n = 17) TempR (n = 10) t (c2) Value* p Value* t (c2) Value† p Value† Age, yrs 48.57 (8.65) 53.59 (13.13) 50.30 (15.50) −1.225 0.230 −0.319 0.755 Sex, male/female‡ 5/9 11/6 6/4 2.584 0.108 1.386 0.239 Education level, yrs 9.50 (4.83) 7.94 (3.17) 9.00 (2.50) 1.079 0.289 0.300 0.767 Handedness, lt/rt/ambi‡ 0/14/0 0/17/0 0/10/0 NA NA NA NA Scores of each cognitive domain DST 11.00 (2.26) 8.09 (2.88) 8.38 (3.11) 2.367 0.210 1.927 0.518 Memory test 11.88 (1.55) 4.27 (4.76) 5.63 (4.93) 4.951 0.001§ 3.423 0.059 Visuospatial test 10.63 (1.60) 4.82 (4.40) 4.13 (3.68) 4.027 0.001§ 4.581 0.003§ DSST 11.88 (1.64) 4.09 (4.32) 4.13 (4.70) 5.455 0.001§ 4.400 0.013§ Mapping 9.88 (0.64) 4.82 (2.79) 4.50 (2.88) 5.812 0.001§ 5.155 0.007§ Similarity 10.00 (1.07) 5.00 (3.52) 5.63 (3.70) 4.437 0.005§ 3.212 0.084 Math exam 10.63 (1.99) 5.82 (2.44) 6.50 (2.33) 4.559 0.001§ 3.803 0.013§ Ambi = ambidextrous; DST = Digit Span Test; HC = healthy control; NA = not applicable; TempR = patients with right temporal glioma. Values are expressed as the mean (SD) unless indicated otherwise. All p values were obtained using the t-test except for sex (chi-square test). Comparisons between controls and patients with left temporal glioma were not performed because 4 patients did not complete all cognitive tests. * Two-sample t-tests between all patients with temporal glioma and controls. † Two-sample t-test between patients with right temporal glioma and controls. ‡ Values are number of patients. § Significant differences were found between controls and patients with glioma. Bonferroni correction for multiple comparisons was performed at a significance level of p < 0.05 (7 tests). regional activity compensation was specific to the contra- creased ALFF value and potential reorganizational time lateral homotopic region. When we ignored the side but (r = 0.335, p = 0.258) for all patients in this group and for compared contralateral ALFF changes in low-grade ver- patients with right temporal lobe LGG (n = 7; r = 0.591, p sus high-grade tumors, patients with LGG (n = 7) in the = 0.204). right temporal lobe exhibited similar hyperactivity in contralateral homotopic regions, which did not present in pa- Discussion tients with left temporal lobe HGGs (n = 6) (Supplemental Data Figs. S2–S4 and Table S2). In this study, we combined ALFF and FC analyses to evaluate possible functional compensation in patients with Reorganization of Global FC in Patients With Right temporal glioma. ALFF was significantly increased in the Temporal Lobe Glioma left hippocampus and parahippocampal gyrus in patients with right temporal glioma, indicating homotopic func- In patients with right temporal lobe glioma, the in- tional plasticity of the contralesional temporal lobe. Our creased ALFF in the left hippocampus and parahippo- study provides evidence of a homotopic functional reorga- campal gyrus (Fig. 1B) was selected as the ROI. Com- nization in patients with unilateral temporal glioma. pared with controls, this group exhibited significantly increased FC to the left inferior temporal gyrus (voxel p Functional Plasticity of the Contralateral Homotopic < 0.0005, GRF corrected cluster p < 0.05). In addition, Region in Patients With Right Temporal Lobe Glioma this group showed significantly decreased FC to the left The brain is an entity capable of meeting various physi- inferior frontal gyrus (voxel p < 0.0005, GRF corrected ological and pathological demands and can effectively cluster p < 0.05). The details of the regions are summa- circumvent the expected functional deterioration of struc- rized in Table 2. tural impairment caused by invasive tumors.6,14 The development of a lesion can affect the structural reorganiza- Correlation Between the Altered Intrinsic Regional Activity tion and compensation of brain function following tumor and Clinical Characteristics growth.1,7,13 Spatiotemporal functional compensation may In patients with right temporal lobe glioma, Pearson’s explain why there is no observable clinical deficit despite partial correlation analysis after controlling for the effects tumor infiltration in the eloquent regions before and after of age, sex, and years of education revealed that the mean surgery.7 The contralesional functional compensation seen ALFF value was significantly negatively correlated with in patients with right temporal lobe glioma is consistent visuospatial scores (r = -0.718, p = 0.045). Age, sex, and with that in a previous voxel-based morphometry structur- years of education were considered as covariates (Fig. al study that showed that infiltration of the unilateral in- 2). No significant correlation was observed between the sula with slow-growing glioma significantly increased the ALFF/FC values and Digital Span Test, DSST, memo- contralateral gray matter volume. Furthermore, the struc- ry, mapping, similarity, or math test scores. In addition, tural plasticity, intrinsic neural remodeling, and network no significant correlation was observed between the in- reorganization in the hippocampus and parahippocampal

420 J Neurosurg Volume 134 • February 2021

Unauthenticated | Downloaded 09/29/21 07:53 AM UTC Liu et al.

FIG. 1. A: Overlap maps of gliomas in patients with right temporal glioma. The warm colors indicate regions, the color bar represents the number of patients with a right temporal lobe glioma on each voxel, and light purple indicates the homotopic mask in contralesional left hemisphere. B: Red indicates increased ALFF within the homotopic mask in patients with right temporal glioma (permutation test, TFCE-FWE corrected p < 0.05), which was selected as the ROI. C: ROI-based whole-brain FC. Red and blue indicate increased and decreased FC, respectively (voxel level p < 0.0005 and cluster-level p < 0.05, GRF corrected). D: ALFF in patients with right temporal glioma (TempR) compared with healthy controls (HC) in the homotopic mask (left, light blue) and 2 control ROIs (left medial frontal ROI [green] and left thalamus ROI [blue]). The histogram represents the mean standardized ALFF extracted from the region showing between-group differences in ALFF analysis and the 2 control ROIs, respectively. Bars and error bars represent the mean ± SEM. ***p < 0.001. Hip/ParaHip = hippocampus and parahippocampal gyrus; NS = not significant. Figure is available in color online only. gyrus have been demonstrated in several studies.8,18,40 Sid- not exhibit a similar functional compensation in the con- hu et al. also detected increased left anterior hippocampal tralesional homotopic regions (i.e., the right hippocampus activation in patients with right temporal lobe epilepsy af- and parahippocampal gyrus) as we expected. We surmise ter anterior temporal lobe resection.33 Consistent with this, that this result, besides the small number of patients, might another study showed that patients with unilateral post- be due to the low proportion of patients with LGGs (n = 1) stroke aphasia have overactive ALFF in the contralesional in this subgroup. In pathological conditions, studies have mesial temporal hippocampus and parahippocampal gy- shown that contrahemispheric plasticity is more effective rus and the lateral temporal cortices.40 These phenomena in progressive rather than extensive and acute injuries.1,14 could be due to the fact that terminal arborizations with Consistent with this, in our research, 7 of the 10 patients high plasticity properties account for maximum synaptic with right temporal lobe glioma had an LGG (WHO grade turnover of adult mossy fibers in the hippocampus,8 which I or II), whereas only 1 of the 7 patients with left tempo- might have contributed to a higher functional plastic poral glioma had an LGG. Our results provide support for tential in the contralesional region in patients with right the functional plasticity of the homotopic hippocampus temporal lobe glioma. In summary, these findings imply and parahippocampal gyrus in patients with LGG. This that highly functional plasticity in the contralateral ho- is clinically significant when considering the resection motopic hippocampus and parahippocampal gyrus might of infiltrative tumors and may help in the evaluation and contribute to maintaining functional balance in patients development of more effective treatment or rehabilitation with unilateral brain lesions. This observation might have strategies. Some technology, such as transcranial magnet- clinical significance when considering the preoperative ic stimulation28 or deep brain stimulation,35 was previously evaluation and iterative resection of gliomas from elo- reported as having the ability to inhibit or excite neural quent areas of the brain.29 structures. By stimulating the contralateral brain region, Interestingly, patients with left temporal lobe glioma did we can evaluate the possible corresponding clinical conse-

J Neurosurg Volume 134 • February 2021 421

Unauthenticated | Downloaded 09/29/21 07:53 AM UTC Liu et al.

TABLE 2. Details of the regions showing a significant difference in ALFF or FC between the temporal glioma group and controls

Peak MNI Cluster Size Mean Strength† TempR vs Controls Region Coordinate* (voxels) TempR HCs t Value Corrected p Value ALFF TempR > HCs Lt Hip & ParaHip gyrus (−33, −15, −21) 33 −0.33 (0.19) −0.72 (0.17) 5.4103 <0.05‡ Functional connectivity TempR < HCs Lt inferior frontal gyrus (−48, 9, 21) 140 −0.10 (0.09) 0.21 (0.14) −5.762 <0.05§ TempR > HCs Lt inferior temporal gyrus (−51, −12, −36) 46 0.59 (0.17) 0.19 (0.19) 6.018 <0.05§ Hip = hippocampus; ParaHip = parahippocampal gyrus. * Peak coordinates (x, y, z) of the MNI brain atlas. † Expressed as the mean z score (SD). ‡ Corrected by TFCE-FWE correction (1000 times of permutation test). § Corrected by GRF correction (voxel-level p < 0.0005, cluster-level p < 0.05). quences before the first operation or speed up the recovery manifested as enhanced FC. On the other hand, this over- of certain functions after resection and cognitive rehabili- activity indicates a compensatory relationship between the tation, and before the second operation. unilateral hippocampus and parahippocampal gyrus and In addition, we supposed that correlation analysis might the inferior temporal gyrus due to the increased demands reveal the relationship between time and fMRI changes of normal processes (such as memory ability). The recruit- when we ignored grade but correlated time with the en- ment of homologous functional compensation–related re- hanced ALFF in patients with right temporal lobe glioma gions in the contralesional hemisphere might contribute to (n = 10). Contrary to prediction, there was no significant shifting the balance of the endogenous activity affected by correlation between the increased ALFF value and po- unilateral lesions during the resting state. tential reorganization time in patients with right temporal Compared with controls, decreased resting-state FC lobe glioma (n = 10), which held true for the LGG patients was observed in the contralesional inferior frontal gyrus, only (n = 7). One possibility is that the small sample size a region remote to the lesion or the homologous mask, in might have concealed some potential relationship. A sec- patients with right temporal lobe glioma. To the best of our ond possibility is that the real compensation time, rather knowledge, the regions with decreased FC (Brodmann ar- than the time from a patient’s first onset of symptoms, can- eas 6, 44, and 45) and the region within major lesion over- not be estimated or determined accurately. Future longi- lap (Brodmann areas 37 and 40) in patients with right tem- tudinal studies focusing on the fMRI changes and their poral lobe glioma are part of the visuospatial network.31,32 relationship with accurate compensational time will better Studies showed that unilateral glioma may cause varying delineate the neuroplasticity in patients with glioma. degrees of alterations in intrinsic FC, including local and long-range effects on the intra- and cross-hemisphere in- Alteration of ROI-Based Whole-Brain FC teractions.2,42 Significantly decreased FC might therefore We defined the region showing significant between- indicate a disruption of network interactions between the group differences in ALFF analysis for ROI-based voxel- behavior-specific network (i.e., visuospatial in this study) level FC analysis (only in nonlesional regions). We found in patients with right temporal glioma. two regions with completely opposite resting-state FC changes between patients with right temporal lobe glioma Correlation Between Intrinsic Regional Activity and and controls. Complete interpretation of these changes in Neurocognitive Assessments the intrinsic resting-state neural activity requires not only Our findings revealed the behavioral significance of regional but also network-based perspectives. With focus contralesional homotopic compensation in patients with on the hyperconnectivity in the left inferior temporal gy- right temporal lobe glioma. The hippocampus and parahip- rus (Fig. 1C, red cluster), our results indicate that the re- pocampal gyrus are involved in the memory circuit,15 and gion with increased FC might play a similar role in this Sidhu et al. reported that the contralateral hippocampus contralesional functional compensatory process. Even contributes to memory recovery in patients who underwent though some differences might exist, our observation was unilateral anterior temporal lobe resection.33 Interestingly, similar to that in a previous study, which showed that re- the imaging-behavioral correlation within these regions mote cerebellar regions with decreased ALFF exhibited was negative in our study, rather than positive as routinely increased FC in the contralesional region in patients with seen. Indeed, when evaluating neurocognitive character- LGG.43 Here, we propose 2 alternative explanations. One istics (Table 1), our results showed that all patients with possible reason for this discrepancy is that we did not take temporal glioma (both left and right) performed worse peripheral edema and the possible effects of tumor infiltra- than the controls on the memory test. However, subgroup tion into account, which is also crucial in neural plastici- analysis showed no significant difference in the memory ty. 5,17 As a result, the lesion overlay map and the homotopic scores between patients with right temporal lobe glioma mask did not contain the inferior temporal gyrus region, and controls. The only behavioral manifestation signifi- and the compensatory overactivity in the homotopic region cantly (negatively) correlated within this region was vi-

422 J Neurosurg Volume 134 • February 2021

Unauthenticated | Downloaded 09/29/21 07:53 AM UTC Liu et al.

could have provided us with more information and should be adopted in future studies. Third, despite the macrostruc- tural evidence that bilateral contralesional plasticity might be a physiological basis for functional compensation,1 we still have to realize that the subjects included in this study were all right-hand dominant. Therefore, we were unable to investigate whether a nondominant temporal glioma would lead to a greater level of plasticity in the unaffected dominant side. Further studies with larger samples of patients with left temporal glioma or left-handed patients are required to verify the role of handedness on hemispheric plasticity. Fourth, many factors can contribute to the functional remodeling detected by fMRI, such as neuronal or glial cell genesis, or even blood flow,25 which were not investigated in our study. The combination of other tech- nologies, such as diffusion tensor imaging and electrical stimulation–based tools, may provide valuable insights FIG. 2. Correlation between ALFF and clinical scores in patients with into the functional, structural, and neurobiological basis right temporal glioma (triangles). Two patients did not complete the visuospatial test. Standardized ALFF values in the left hippocampus and of this compensatory process. Last but not least, we have parahippocampal region were negatively correlated with visuospatial to bear in mind that the clinical significance of our imag- scores in patients with right temporal glioma (r = –0.718, p = 0.045). The ing findings could not be either absolutely summarized or solid line and dashed lines represent the best-fit line and 95% confi- excluded from the specific cognitive domains that we have dence interval of the Pearson partial correlation, respectively. Figure is evaluated in our study. More comprehensive assessments available in color online only. of multidomain cognitive functions are needed to explain the clinical significance of the imaging findings at the -in dividual level during this process. suospatial ability. Consistent with our findings, Yang et al. reported that the enhanced ALFFs in the contralesional Conclusions hippocampus and parahippocampal gyrus were negatively Our results provide evidence for a homotopic function- correlated with the behavioral scores.40 We surmised that, al reorganization of the left hippocampus and parahippo- although increased intrinsic neural activity in the homo- campal gyrus in patients with right temporal glioma (no topic region (i.e., left hippocampus and parahippocampal positive findings in patients with left temporal lobe glioma gyrus) might reflect—or at least tend to reflect—memory- due to small sample size or small proportion of LGG pa- related functional compensation, this compensation did tients), which may contribute to memory recovery and pre- not contribute to the visuospatial abilities observed in the dict the extent of visuospatial impairment in patients with partial correlation analysis. These findings indicated that right temporal glioma. Moreover, the homotopic functional high intrinsic regional neural activity in the contralesional alteration might be involved in potential various network hippocampus and parahippocampal gyrus was associated interactions during such remodeling processes. Our find- with the high clinical deficit caused by unilateral lesions. ings provide novel insights into brain plasticity in patients The damaged visuospatial function in patients with right with unilateral glioma. temporal lobe glioma was the result of the damage caused by lesions in that region. Consequently, the slow progress Acknowledgments of lesion development prolonged the time of the homotopic This study was supported by a grant from the clinical medical functional compensation, which also worsened the visuo- scientific and technologic project of National Health and Family spatial abilities. Therefore, we hypothesize that increased Planning Commission of the People’s Republic of China (grant ALFF in contralesional homotopic regions is predictive of no. w201308), a grant from the project of Jiangsu Provincial functional reorganization in patients with right temporal Medical Youth Talent (grant no. QNRC2016047), the Nanjing lobe glioma. Commission of Health and Family Planning (grant no. H201540), a grant from the medical scientific and technologic development Limitations and Future Directions project of Nanjing (grant no. YKK12137 and ZKX15035), and a grant from the project of Jiangsu Provincial Medical Innovation There are several limitations in the current study that Team (grant no. CXTDA2017050). need to be addressed. First, the small sample size could have resulted in bias and obviated statistical significance in the ALFF values between in patients with left tempo- References ral lobe glioma and controls. This also prevented an ROI- 1. Almairac F, Duffau H, Herbet G: Contralesional macrostruc- based FC analysis in these patients. It is possible there- tural plasticity of the insular cortex in patients with glioma: a VBM study. Neurology 91:e1902–e1908, 2018 fore that the contralesional reorganizational ability differs 2. Alstott J, Breakspear M, Hagmann P, Cammoun L, Sporns between LGG and HGG. Second, the new 2016 WHO O: Modeling the impact of lesions in the human brain. PLoS classification system incorporates molecular biomarkers, Comput Biol 5:e1000408, 2009 together with classic histological features in an integrated 3. Chen X, Lu B, Yan CG: Reproducibility of R-fMRI metrics diagnosis. The accurate molecular subtypes of gliomas on the impact of different strategies for multiple comparison

J Neurosurg Volume 134 • February 2021 423

Unauthenticated | Downloaded 09/29/21 07:53 AM UTC Liu et al.

correction and sample sizes. Hum Brain Mapp 39:300–318, neural activity in the sensorimotor system: a resting-state 2018 fMRI study. Neural Plast 2018:2024835, 2018 4. Derks J, Dirkson AR, de Witt Hamer PC, van Geest Q, Hulst 24. Maesawa S, Bagarinao E, Fujii M, Futamura M, Motomura HE, Barkhof F, et al: Connectomic profile and clinical pheno- K, Watanabe H, et al: Evaluation of resting state networks type in newly diagnosed glioma patients. Neuroimage Clin in patients with gliomas: connectivity changes in the unaf- 14:87–96, 2017 fected side and its relation to cognitive function. PLoS One 5. Duffau H: The huge plastic potential of adult brain and the 10:e0118072, 2015 role of connectomics: new insights provided by serial map- 25. May A, Hajak G, Gänssbauer S, Steffens T, Langguth B, pings in glioma surgery. Cortex 58:325–337, 2014 Kleinjung T, et al: Structural brain alterations following 5 6. Duffau H: Lessons from brain mapping in surgery for low- days of intervention: dynamic aspects of neuroplasticity. grade glioma: insights into associations between tumour and Cereb Cortex 17:205–210, 2007 brain plasticity. Lancet Neurol 4:476–486, 2005 26. Ostrom QT, Bauchet L, Davis FG, Deltour I, Fisher JL, 7. Duffau H, Capelle L, Denvil D, Sichez N, Gatignol P, Lopes Langer CE, et al: The epidemiology of glioma in adults: a M, et al: Functional recovery after surgical resection of low “state of the science” review. Neuro Oncol 16:896–913, 2014 grade gliomas in eloquent brain: hypothesis of brain compen- 27. Parisot S, Darlix A, Baumann C, Zouaoui S, Yordanova Y, sation. J Neurol Neurosurg Psychiatry 74:901–907, 2003 Blonski M, et al: A probabilistic atlas of diffuse WHO grade 8. Galimberti I, Bednarek E, Donato F, Caroni P: EphA4 sig- II glioma locations in the brain. PLoS One 11:e0144200, naling in juveniles establishes topographic specificity of 2016 structural plasticity in the hippocampus. Neuron 65:627– 28. Pascual-Leone A, Tarazona F, Keenan J, Tormos JM, Ham- 642, 2010 ilton R, Catala MD: Transcranial magnetic stimulation and 9. Gallea C, Popa T, García-Lorenzo D, Valabregue R, Legrand neuroplasticity. Neuropsychologia 37:207–217, 1999 AP, Apartis E, et al: Orthostatic tremor: a cerebellar pathol- 29. Picart T, Herbet G, Moritz-Gasser S, Duffau H: Iterative sur- ogy? Brain 139:2182–2197, 2016 gical resections of diffuse glioma with awake mapping: how 10. Gauthier LV, Taub E, Perkins C, Ortmann M, Mark VW, to deal with cortical plasticity and connectomal constraints? Uswatte G: Remodeling the brain: plastic structural brain Neurosurgery 85:105–116, 2019 changes produced by different motor therapies after stroke. 30. Raysi Dehcordi S, Mariano M, Mazza M, Galzio RJ: Cogni- Stroke 39:1520–1525, 2008 tive deficits in patients with low and high grade gliomas. J 11. Greicius M: Resting-state functional connectivity in neuro- Neurosurg Sci 57:259–266, 2013 psychiatric disorders. Curr Opin Neurol 21:424–430, 2008 31. Richiardi J, Altmann A, Milazzo AC, Chang C, Chakravarty 12. Gupta L, Gupta RK, Postma AA, Sahoo P, Gupta PK, Patir MM, Banaschewski T, et al: Correlated gene expression R, et al: Advanced and amplified BOLD fluctuations in high- supports synchronous activity in brain networks. Science grade gliomas. J Magn Reson Imaging 47:1616–1625, 2018 348:1241–1244, 2015 13. Heiss WD, Thiel A, Kessler J, Herholz K: Disturbance and 32. Shirer WR, Ryali S, Rykhlevskaia E, Menon V, Greicius MD: recovery of language function: correlates in PET activation Decoding subject-driven cognitive states with whole-brain studies. Neuroimage 20 (Suppl 1):S42–S49, 2003 connectivity patterns. Cereb Cortex 22:158–165, 2012 14. Herbet G, Maheu M, Costi E, Lafargue G, Duffau H: Map- 33. Sidhu MK, Stretton J, Winston GP, McEvoy AW, Symms M, ping neuroplastic potential in brain-damaged patients. Brain Thompson PJ, et al: Memory network plasticity after tempo- 139:829–844, 2016 ral lobe resection: a longitudinal functional imaging study. 15. Jeffery KJ: The hippocampus: from memory, to map, to Brain 139:415–430, 2016 memory map. Trends Neurosci 41:64–66, 2018 34. Smith SM, Nichols TE: Threshold-free cluster enhancement: 16. Kinno R, Ohta S, Muragaki Y, Maruyama T, Sakai KL: Dif- addressing problems of smoothing, threshold dependence ferential reorganization of three syntax-related networks and localisation in cluster inference. Neuroimage 44:83–98, induced by a left frontal glioma. Brain 137:1193–1212, 2014 2009 17. Kristo G, Raemaekers M, Rutten GJ, de Gelder B, Ramsey 35. Udupa K, Bahl N, Ni Z, Gunraj C, Mazzella F, Moro E, et al: NF: Inter-hemispheric language functional reorganization Cortical plasticity induction by pairing subthalamic nucleus in low-grade glioma patients after tumour surgery. Cortex deep-brain stimulation and primary motor cortical transcra- 64:235–248, 2015 nial magnetic stimulation in Parkinson’s Disease. J Neurosci 18. Li H, Ji C, Zhu L, Huang P, Jiang B, Xu X, et al: Reorgani- 36:396–404, 2016 zation of anterior and posterior hippocampal networks as- 36. Volz LJ, Rehme AK, Michely J, Nettekoven C, Eickhoff SB, sociated with memory performance in mesial temporal lobe Fink GR, et al: Shaping early reorganization of neural net- epilepsy. Clin Neurophysiol 128:830–838, 2017 works promotes motor function after stroke. Cereb Cortex 19. Li Y, Tan Z, Wang J, Wang Y, Gan Y, Wen F, et al: Altera- 26:2882–2894, 2016 tions in spontaneous brain activity and functional network 37. Voytek B, Davis M, Yago E, Barceló F, Vogel EK, Knight reorganization following surgery in children with medically RT: Dynamic neuroplasticity after human prefrontal cortex refractory epilepsy: a resting-state functional magnetic reso- damage. Neuron 68:401–408, 2010 nance imaging study. Front Neurol 8:374, 2017 38. Whitcroft KL, Fischer J, Han P, Raue C, Bensafi M, Gudziol 20. Liu D, Hu X, Liu Y, Yang K, Xiao C, Hu J, et al: Potential V, et al: Structural plasticity of the primary and secondary intra- or cross-network functional reorganization of the triple olfactory cortices: increased gray matter volume following unifying networks in patients with frontal glioma. World surgical treatment for chronic rhinosinusitis. Neuroscience Neurosurg 128:e732–e743, 2019 395:22–34, 2018 21. Liu G, Dang C, Peng K, Xie C, Chen H, Xing S, et al: 39. Yan CG, Wang XD, Zuo XN, Zang YF: DPABI: Data pro- Increased spontaneous neuronal activity in structurally cessing & analysis for (resting-state) brain imaging. Neuro- damaged cortex is correlated with early motor recovery in informatics 14:339–351, 2016 patients with subcortical infarction. Eur J Neurol 22:1540– 40. Yang M, Li J, Li Y, Li R, Pang Y, Yao D, et al: Altered intrin- 1547, 2015 sic regional activity and interregional functional connectivity 22. Lopour BA, Tavassoli A, Fried I, Ringach DL: Coding of in post-stroke aphasia. Sci Rep 6:24803, 2016 information in the phase of local field potentials within hu- 41. Zang Y, Jiang T, Lu Y, He Y, Tian L: Regional homogeneity man medial temporal lobe. Neuron 79:594–606, 2013 approach to fMRI data analysis. Neuroimage 22:394–400, 23. Lu Y, Zhao Q, Wang Y, Zhou C: Ballroom dancing promotes 2004

424 J Neurosurg Volume 134 • February 2021

Unauthenticated | Downloaded 09/29/21 07:53 AM UTC Liu et al.

42. Zhang H, Shi Y, Yao C, Tang W, Yao D, Zhang C, et al: the article: H Liu, D Liu, Chen, X Hu. Critically revising the arti- Alteration of the intra- and cross-hemisphere posterior de- cle: H Liu, Chen, X Hu, G Hu, Y Liu, Yang, Xiao, Zou. Reviewed fault mode network in frontal lobe glioma patients. Sci Rep submitted version of manuscript: H Liu, X Hu, G Hu, Y Liu, 6:26972, 2016 Yang, Xiao, Zou. Approved the final version of the manuscript 43. Zhang N, Xia M, Qiu T, Wang X, Lin CP, Guo Q, et al: Reor- on behalf of all authors: H Liu. Statistical analysis: H Liu, D Liu, ganization of cerebro-cerebellar circuit in patients with left G Hu. Administrative/technical/material support: H Liu. Study hemispheric gliomas involving language network: A com- supervision: H Liu, Zou. bined structural and resting-state functional MRI study. Hum Brain Mapp 39:4802–4819, 2018 Supplemental Information Online-Only Content Disclosures Supplemental material is available with the online version of the article. The authors report no conflict of interest concerning the materi- Supplemental Data. https://thejns.org/doi/suppl/10.3171/2019. als or methods used in this study or the findings specified in this 11.JNS191982. paper. Correspondence Author Contributions Hongyi Liu: The Affiliated Brain Hospital of Nanjing Medical Conception and design: H Liu, D Liu, Chen. Acquisition of data: University, Jiangsu, China. [email protected]. D Liu, Chen, X Hu, G Hu, Y Liu, Yang, Xiao, Zou. Analysis and interpretation of data: D Liu, Chen, X Hu, Y Liu, Zou. Drafting

J Neurosurg Volume 134 • February 2021 425

Unauthenticated | Downloaded 09/29/21 07:53 AM UTC